Applying SOLID principles to your TypeScript code

栏目: IT技术 · 发布时间: 4年前

内容简介:The SOLID principles were defined quite some time ago and are still relatable. Their goal is to make our software easier to understand, read, and extend. We accredit this concept toThe single responsibility principle declares that a class should only be re

The SOLID principles were defined quite some time ago and are still relatable. Their goal is to make our software easier to understand, read, and extend. We accredit this concept to Robert C. Martin from his paper from the year 2000. The actual SOLID acronym was defined later, though. In this article, we go through all of the SOLID principles and reflect them in TypeScript examples.

Single responsibility principle

The single responsibility principle declares that a class should only be responsible for a single functionality. The above also refers to modules and functions.

class Statistics {
  public computeSalesStatistics() {
    // ...
  }
  public generateReport() {
    // ...
  }
}

We should split the above class into two separate ones. SOLID defines the concept of responsibility as a reason to change .

Robert C. Martin, in his blogpost , discusses how we can define a reason to change. He stresses that it shouldn’t be deliberated in terms of code. For example, refactoring is not a reason to change in terms of the single responsibility principle.

To define a reason to change, we need to investigate what is the responsibility of our program.

The Statistics class might be changed due to two different reasons:

  • the logic of the sales statistic computation changes,
  • the format of the report changes

The single responsibility principle highlights that the two above aspects put two different responsibilities on the Statistics class.

class Statistics {
  public computeSalesStatistics() {
    // ...
  }
}
class ReportGenerator {
  public generateReport() {
    // ...
  }
}

Now we have to separate classes, and each of them has a single reason to change, and therefore just one responsibility. Applying to the above principle makes our code easier to explain and understand.

Open-closed principle

According to the open-closed principle, software entities should be open for extension but closed for modification .

The core idea of the above principle is that we should be able to add new functionalities without changing the existing code.

class Rectangle {
  public width: number;
  public height: number;
  constructor(width: number, height: number) {
    this.width = width;
    this.height = height;
  }
}
class Circle {
  public radius: number;
  constructor(radius: number) {
    this.radius = radius;
  }
}

Let’s say we want to create a function that calculates the area of an array of shapes. With our current design, it might look like that:

function calculateAreasOfMultipleShapes(
  shapes: Array<Rectangle | Circle>
) {
  return shapes.reduce(
    (calculatedArea, shape) => {
      if (shape instanceof Rectangle) {
        return calculatedArea + shape.width * shape.height;
      }
      if (shape instanceof Circle) {
        return calculatedArea + shape.radius * Math.PI;
      }
    },
    0
  );
}

TypeScript understand what properties our shapes have thanks to control-flow based narrowing.

We discuss that subject a bit in  Understanding any and unknown in TypeScript. Difference between never and void

The issue with the above approach is that when we introduce a new shape, we need to modify our calculateAreasOfMultipleShapes function. This makes it open for modification and breaks the open-closed principle.

We can fix that by enforcing our shapes to have a method that returns the area.

interface Shape {
  getArea(): number;
}
class Rectangle implements Shape {
  public width: number;
  public height: number;
  constructor(width: number, height: number) {
    this.width = width;
    this.height = height;
  }
  public getArea() {
    return this.width * this.height;
  }
}
class Circle implements Shape {
  public radius: number;
  constructor(radius: number) {
    this.radius = radius;
  }
  public getArea() {
    return this.radius * Math.PI;
  }
}

Now that we are sure that all of our shapes have the getArea function, we can use it further.

function calculateAreasOfMultipleShapes(
  shapes: Shape[]
) {
  return shapes.reduce(
    (calculatedArea, shape) => {
      return calculatedArea + shape.getArea();
    },
    0
  );
}

Now when we introduce a new shape, we don’t need to modify our calculateAreasOfMultipleShapes function. We make it  open for extension but closed for modification .

We could achieve the above functionality also by using an abstract class instead of an interface

Liskov substitution principle

The above rule, introduced by Barbara Liskov , also helps us ensure that changing one area of our system does not break other parts. To make this principle less confusing, we will break it down into multiple parts.

Replacing an instance of a class with its child class should not produce any negative side effects

The first thing we notice in the above principle is that its main focus is class inheritance.

Let’s implement a straightforward and vivid example of how we can break the above principle.

class Employee {
  protected permissions: any = new Set<string>();
 
  public hasPermission(permissionName: string) {
    return this.permissions.has(permissionName);
  }
  public addPermission(permissionName: string) {
    return this.permissions.add(permissionName);
  }
}
class Cashier extends Employee {
  protected permissions: string[] = [];
 
  public addPermission(permissionName: string) {
    this.permissions.push(permissionName);
  }
}
function isPersonAllowedToDeleteProducts(person: Employee) {
  return person.hasPermission('deleteProducts');
}

The above code is very problematic because the implementation of permissions differs in the  Cashier and the  User .

const employee = new Employee();
employee.addPermission('deleteProducts');
isPersonAllowedToDeleteProducts(employee);

The above code works fine, but when we replace an instance of a parent class with its child class, we experience issues.

const cashier = new Cashier();
cashier.addPermission('deleteProducts');
isPersonAllowedToDeleteProducts(cashier);
TypeError: this.permissions.has is not a function

This situation is very vivid and shouldn’t happen in a properly typed TypeScript code. We had to use permissions : any to allow the  Cashier to extend the  User improperly.

Validation conditions

A more refined example is with preconditions and validation.

class Employee {
  protected permissions = new Set<string>();
 
  public addPermission(permissionName: string) {
    return this.permissions.add(permissionName);
  }
}
class Cashier extends Employee {
  public addPermission(permissionName: string) {
    if (permissionName === 'deleteProducts') {;
      throw new Error(
        'Cashier should not be able to delete products!'
      );
    }
    return this.permissions.add(permissionName);
  }
}

The above example, on the other hand, is fine in terms of typings. Unfortunately, it also breaks the Liskov substitution principle.

const employee = new Employee();
employee.addPermission('deleteProducts');
const employee = new Cashier();
employee.addPermission('deleteProducts');
Error: Cashier should not be able to delete products!

The same thing applies to output conditions. If the function that we override returns a value, the child class shouldn’t put additional validation on the output.

Interface segregation principle

The interface segregation principle puts an emphasis on creating smaller and more specific interfaces. Let’s imagine the following situation.

interface Bird {
  fly(): void;
  walk(): void;
}

Above, we have an interface of a bird. We assume that birds can walk and fly. It is straightforward to create an example of such a bird:

class Nightingale implements Bird {
  public fly() {
    /// ...
  }
  public walk() {
    /// ...
  }
}

The above is not always the case, though. The above assumption can be incorrect.

class Kiwi implements Bird {
  public fly() {
    throw new Error('Unfortunately, Kiwi can not fly!');
  }
  public walk() {
    /// ...
  }
}

The interface segregation principle states that  no client should be forced to depend on methods it does not use . By putting too many properties in our interfaces, we risk breaking the above rule.

What we might do instead is to implement smaller interfaces, sometimes referred to as role interfaces .

interface CanWalk {
  walk(): void;
}
interface CanFly {
  fly(): void;
}
class Nightingale implements CanFly, CanWalk {
  public fly() {
    /// ...
  }
  public walk() {
    /// ...
  }
}
class Kiwi implements CanWalk {
  public walk() {
    /// ...
  }
}

By changing our approach to interfaces, we avoid bloating them and make our software easier to maintain.

Dependency inversion principle

The core of the dependency inversion principle is that high-level modules should not depend on the low-level modules . Instead, both of them should depend on abstractions.

Let’s create an example to investigate the above further.

interface Person {
  introduceSelf(): void;
}
class Engineer implements Person {
  public introduceSelf() {
    console.log('I am an engineer');
  }
}
class Musician implements Person {
  public introduceSelf() {
    console.log('I am a musician');
  }
}

The above behavior is elementary and academic, but it is not always the case. If the introduction was more complicated, we might want to create separate classes just for that.

interface IntroductionService {
  introduce(): void;
}
class EngineerIntroductionService implements IntroductionService {
  public introduce() {
    console.log('I am an engineer');
  }
}
class Engineer implements Person {
  private introductionService = new EngineerIntroductionService();
  public introduceSelf() {
    this.introductionService.introduce();
  }
}

Unfortunately, the above code breaks the dependency inversion principle. It says that we should invert the dependencies of the Engineer and the  EngineerIntroductionService .

class Engineer implements Person {
  public introductionService: EngineerIntroductionService;
 
  constructor(introductionService: IntroductionService) {
    this.introductionService = introductionService;
  }
 
  public introduceSelf() {
    this.introductionService.introduce();
  }
}
const engineer = new Engineer(new EngineerIntroductionService());

The benefit of the above is that we don’t need subclasses for the Engineer and the Musician .

class Person {
  public introductionService: IntroductionService;
 
  constructor(introductionService: IntroductionService) {
    this.introductionService = introductionService;
  }
  public introduceSelf() {
    this.introductionService.introduce();
  }
}
const engineer = new Person(new EngineerIntroductionService());
const musician = new Person(new MusicianIntroductionService());

The above approach is an example of using composition instead of inheritance .

Also, this makes our classes easier to unit-test, because we can effortlessly provide a mocked service in a constructor.

Summary

In this article, we’ve gone through all of the SOLID principles:

  • S : Single responsibility principle
  • O : Open-closed principle
  • L : Liskov substitution principle
  • I : Interface segregation principle
  • D : Dependency inversion principle

By applying them to our code, we can make it more readable and maintainable. Also, those are good practices that make our codebase easier to expand while not affecting other parts of our application.

Even though SOLID was defined quite some time ago, it shows that it is still relatable, and we can gain some real benefits from understanding them.


以上所述就是小编给大家介绍的《Applying SOLID principles to your TypeScript code》,希望对大家有所帮助,如果大家有任何疑问请给我留言,小编会及时回复大家的。在此也非常感谢大家对 码农网 的支持!

查看所有标签

猜你喜欢:

本站部分资源来源于网络,本站转载出于传递更多信息之目的,版权归原作者或者来源机构所有,如转载稿涉及版权问题,请联系我们

流量池

流量池

杨飞 / 中信出版集团 / 2018-4 / 68.00

移动互联网时代,信息日益冗余,新闻速朽; 整体流量增长速度放缓,而竞争者数量高速增加; 流量呈现变少、变贵、欺诈频繁的现状; 品效合一的营销策略成为共识,而实现路径成为痛点; 多次开创各营销渠道效果之最的营销人、各种刷屏级营销事件操盘手、神州专车CMO杨飞,这一次倾囊相授,诚恳讲述如何实现流量获取、营销转化以及流量的运营和再挖掘。一起来看看 《流量池》 这本书的介绍吧!

JS 压缩/解压工具
JS 压缩/解压工具

在线压缩/解压 JS 代码

随机密码生成器
随机密码生成器

多种字符组合密码

正则表达式在线测试
正则表达式在线测试

正则表达式在线测试